AnyStory: Towards Unified Single and Multiple Subject Personalization in Text-to-Image Generation
V2.0.0
Official demo of AnyStory (FLUX.1-dev version). The code has been released on GitHub.
🚀🚀🚀 Quick Start:
1. Upload reference images for the subjects (clean background; real human IDs unsupported for now), enter text prompts, and click "RUN".
2. (Recommended) Click "Segment Subject" to create masks (or upload your own B&W masks) for subjects. This helps the model better reference the subject you specify (otherwise, we will perform automatic detection).
💡💡💡 Tips:
If the subject doesn't appear, try adding a detailed description of the subject in the prompt that matches the reference image, and avoid conflicting details (e.g., significantly altering the subject's appearance).
1. Upload reference images for the subjects (clean background; real human IDs unsupported for now), enter text prompts, and click "RUN".
2. (Recommended) Click "Segment Subject" to create masks (or upload your own B&W masks) for subjects. This helps the model better reference the subject you specify (otherwise, we will perform automatic detection).
💡💡💡 Tips:
If the subject doesn't appear, try adding a detailed description of the subject in the prompt that matches the reference image, and avoid conflicting details (e.g., significantly altering the subject's appearance).
Examples
| Subject [A] Reference Image | Subject [A] Mask (upload supported) | Subject [B] Reference Image | Subject [B] Mask (upload supported) | Prompt | Generated Image |
|---|
More examples: Intelligent creation of AI story pictures integrated with Qwen Agent
📝 Citation
If our work is helpful for your research or applications, please cite us via:
@article{he2025anystory,
title={AnyStory: Towards Unified Single and Multiple Subject Personalization in Text-to-Image Generation},
author={He, Junjie and Tuo, Yuxiang and Chen, Binghui and Zhong, Chongyang and Geng, Yifeng and Bo, Liefeng},
journal={arXiv preprint arXiv:2501.09503},
year={2025}
}
If you have any questions, feel free to open an issue or contact us directly at hejunjie1103@gmail.com.